Audio Compass

[robkwhitehouse]

I develop gadgets for Visually Impaired (VI) people. A local charity takes VI people out on sailboats. It would be cool to have a repeating audio announcement of the current compass heading on a bluetooth speaker in the boat cockpit. I have already developed the electronic compass using an ESP32 (and an IMU module). I'm hoping to record a set of audio streams using Audacity and save these into the ESP32 flash. Then use BluetoothA2DPSource to transmit the appropriate stream(s) from flash. This would be repeated at a configurable interval e.g. every 10 seconds.

Does all this sound feasible/sensible?
Do I need to use the AudioTools library ?

Many thanks

[pschatzmann]

Here is a simple exampe using the arduino-audio-tools, arudino-simple-tts and this A2DP library

You will just need to build your own dictionary that includes the four cardinal directions...

[robkwhitehouse]

Thanks! In fact I want the compass heading read out as a 3 digit number e.g. 059 degrees would be "OH FIVE NINE". So after a quick look at your Audio Tools it seems that you have already provided most of the classes that I need. I will have a play.

Thanks again

[robkwhitehouse]

Phil,

I have now managed to get the arduino-simple-tts example to build. It was a slight struggle because there is another dependency not mentioned in the README - i.e. Arduino-libhelix also need to be installed.

[robkwhitehouse]

Phil,

I have been have a good look through your TTS options. I'm currently thinking that listening to headings announced using any of the synthesised TTS libraries are going to get rather tiresome on a long voyage. So I will stick with the AudioDictionary idea. I will use Audacity to record 360 mp3 files. One for each possible compass heading. I think that If I record these in mono at 8khz sample rate then I should just about be able to squeeze them into the ESP32 PROGMEM. Any tips on converting mp3 files to the ASCII hex?
I will also modify the AudioDictionaryEntry struct to remove the index string. I will simply load the mp3 data into an array e.g. AudioDictonaryEntry[360] so that I can access the elements without searching (to reduce latency).

Does this make any sense?

Thanks,

Bob

[pschatzmann]

I don't think that the change of latency will be noticable...

Please check the Wiki: Generation of Audio Files in the simple-tts project...

[pschatzmann]

With just 4 recordings you should be able to generate all the directions mentioned above...

[robkwhitehouse]

Thanks. I could use the cardinal points, as above, but people steering a boat generally like to have a more fine grained heading to steer. So a numeric announcement of the current heading in degrees is good. People can rarely steer that accurately but they like to think that they can!